PR measurement should replace advertising value equivalency (AVE) with outcomes a CFO accepts: referral visits, key events, share of AI answers and competitor sentiment. AMEC's Barcelona Principles 3.0, issued in 2020, state that AVEs are not the value of communication, and AMEC's 2020 Global Member Survey found only 3% of practitioners call AVEs relevant to an evaluation program, as reported by CARMA. This guide shows how to build the replacement in GA4 and in AI answers.
What Is AVE and Why Do CFOs Reject It?
Advertising value equivalency (AVE) is the price of buying the same space or airtime as a piece of editorial coverage, and CFOs reject it because it prices attention instead of results. The Open University Business School describes AVE as what it would have cost to buy the space that coverage took up.
The flaw is that AVE counts size, not effect. AMEC's guidance, as summarized by BI Norwegian Business School, warns that AVEs only account for the size of coverage and equate cost with value. A large story that sends no qualified visitor to the website earns a high AVE and zero pipeline.
A CFO asks a different question: what did this spend change in revenue, cost or risk? AVE has no answer, so the finance team treats the whole communications report as unproven. The fix is to report measurements that connect to a business outcome and to state the goal before the campaign starts. The Barcelona Principles put goal setting first, and they call for both qualitative and quantitative analysis, according to PR News.
Which Vanity Metrics Should Boardroom Metrics Replace?
Five vanity metrics should be replaced by boardroom metrics that tie coverage to a visit, a key event or a buyer's view of the brand. The table pairs each one with a measurement a finance team can audit.
| Vanity metric | Why it fails | Boardroom metric |
|---|---|---|
| AVE | Prices coverage like ad space and ignores what readers did next | Key events (demo requests, trial sign-ups) from sessions that began with earned coverage |
| Impressions and reach | Counts potential audience, not action | Referral visits from each outlet and their engagement rate |
| Clip count | Rewards volume, not placement quality | Source share: the share of coverage in outlets that AI engines cite for the category |
| Press release pickups | Press releases make up a small share of AI citations | Placements in outlets cited in AI answers for buyer prompts |
| Overall sentiment score | Has no baseline, so a good score may still trail rivals | Sentiment compared with the top three competitors in AI answers |
The fourth row rests on citation data. Muck Rack's December 2025 report found that earned media made up 82% of AI citations. Press releases were 1.2% of citations in Muck Rack's July 2025 report and 6% in December 2025, according to a summary of its May 2026 edition on the National Newspaper Association site. A press release pickup is therefore a weak proxy for the sources buyers' chatbots read.
How Do You Connect Each Goal to a Metric Before Reporting?
Each communications goal needs one metric, one data source and one owner before the campaign starts, so the report answers the question the CFO asked. The table below maps three common goals to measurements that can be audited.
| Goal | Metric | Data source |
|---|---|---|
| Generate pipeline | Key events from sessions with an earned-media touchpoint | GA4 attribution paths and explorations |
| Be named when buyers ask a chatbot | Prompt share: the share of buyer prompts that name the brand | Quarterly prompt runs across five AI engines |
| Hold or improve reputation against rivals | Sentiment and descriptors versus the top three competitors | The same prompt runs, scored per engine |
The second row connects to buyer behavior. G2's March 2026 survey of 1,076 software buyers found that 85% think more highly of a vendor when an AI chatbot includes it in an answer. A finance team can follow the chain from mention rate to buyer opinion to pipeline.
How Do You Set Up GA4 Referral Paths for Earned Media?
GA4 referral paths for earned media come from three setup steps: mark the outcomes as key events, group earned and AI traffic into their own channels, and read the paths in the attribution and exploration reports. Google's help center notes that conversions are now called key events, and that attribution reports show how credit is spread across the touchpoints before a key event.
| Step | What to do in GA4 | Result |
|---|---|---|
| 1 | Mark demo requests, trial sign-ups and pricing-page visits as key events | Outcomes the CFO already recognizes |
| 2 | Create a custom channel group named "Earned Media" with a rule on session source matching the domains of outlets that covered the brand | Earned referrals separated from the general Referral channel |
| 3 | Create a second channel named "AI Assistants" with a source rule for chatgpt.com, perplexity.ai, gemini.google.com, copilot.microsoft.com and claude.ai | Chatbot visits reported on their own line |
| 4 | Move both channels above Referral in the channel order | Visits counted in the new channels, not in the default bucket |
| 5 | Open the attribution paths report for each key event and filter for paths that include the new channels | The sequence of touchpoints that ends in a key event |
| 6 | Build a path exploration that starts at each coverage landing page | Page-level routes from an article visit to a key event |
Two limits apply. Google states that attribution models exclude direct visits unless the whole path is direct, so a buyer who reads an article and later types the company name gets no credit. Add branded search impressions from Google Search Console to the report to capture that lift.
The second limit is the chatbot. ChatGPT adds a utm_source=chatgpt.com tag to citation links, but Seer Interactive's testing found the tag is not guaranteed on every link. Themeisle's GA4 guide concludes that GA4 undercounts AI traffic, and Shero Commerce notes that visitors who paste a chatbot link into a new tab arrive with no referrer. A "How did you hear about us?" field on the demo form closes part of the gap.
What Does a Share-of-AI-Answers Measurement Look Like?
A share-of-AI-answers measurement is the percentage of buyer prompts in which an AI engine names the brand, and it is the AI-era counterpart of share of voice. It needs a fixed prompt list, a fixed set of engines and a repeat schedule so the numbers can be compared over time.
5W recommends 20 to 30 buyer prompts per category, repeated each quarter, in its accounting software AI Visibility Index. The prompts should cover category, comparison, alternatives and requirements questions. Each prompt runs in ChatGPT, Claude, Perplexity, Gemini and Copilot, and each answer is logged.
Engines must be scored separately. Muck Rack's December 2025 report found that the most-cited outlets shifted across AI models with minimal overlap, so a brand can lead in one engine and be absent from another. One blended number hides that gap.
How Do You Benchmark Competitor Sentiment Inside AI Answers?
Competitor sentiment in AI answers is benchmarked by running the same prompts for the brand and its top three rivals, then scoring how each one is described. The result shows whether the brand is merely mentioned or described as the stronger choice.
| Dimension | What to record | What it tells the CFO |
|---|---|---|
| Mention rate | Whether the brand is named, per prompt and per engine | Whether the brand is on the shortlist at all |
| List position | Where the brand appears among the names given | How strongly the engine favors it |
| Descriptors | The adjectives and claims the answer attaches to the brand | What buyers read about the brand before they contact it |
| Sentiment | Positive, neutral or negative framing, scored by the same rubric for every brand | The brand's standing against each rival |
| Cited sources | The outlets and pages the answer links | Which placements to repeat or fix |
| Competitor gap | The rivals named when the brand is not | Where the next campaign should aim |
History belongs in the benchmark. 5W's Reputation Index Study 08 measured how five AI engines render named people across sixty-plus retrieval-intent prompts per engine. In that study, one past event surfaced in 75% of identity prompts for a single coach despite a successful later tenure. The lesson for a brand is that an old incident can keep appearing in answers long after the news cycle ends, so the benchmark must log which past events each engine repeats.
The Index describes its scores as directional estimates from modeled engine outputs. Treat any sentiment score the same way: report the method, apply it to every brand in the same way, and compare changes between runs, not single readings. For the repair side of the same problem, see 5W's guide to online reputation management services and cost.
How Should You Present PR Results to the CFO?
PR results should reach the CFO as a one-page report that leads with outcomes, shows the change since the last run and states the method. The page replaces the clip book and the AVE total.
The report has four blocks. The first shows key events from sessions that include an earned-media or AI-assistant touchpoint. The second shows prompt share and the competitor gap for the quarter. The third lists the placements in outlets that engines cited, and the fourth states what changes next quarter. Each block cites its data source, so the finance team can verify the numbers.
Keep the language plain. A CFO does not need the term "retrieval layer." The CFO needs to see that buyers who asked a chatbot for vendors named the brand in more answers this quarter than last, and that visits and key events from those paths rose with it.
Pair the report with the media plan. Muck Rack's December 2025 analysis found that the journalists PR teams pitch most overlapped with the journalists AI engines cite most by an average of only 2%. If the pitch list comes from the citation map instead of the usual targets, the next report shows the effect in the same metrics. For a planning example, see how 5W frames enterprise tech PR and AI search as one program.
Frequently Asked Questions
What replaces AVE in PR measurement?
Outcome metrics replace AVE: key events from sessions that include earned coverage, referral visits by outlet, share of AI answers that name the brand, and sentiment against competitors. Each metric needs a stated goal before the campaign starts, in line with the Barcelona Principles.
How do you track ChatGPT referrals in GA4?
Create a custom channel group with a source rule for chatgpt.com and the other AI assistant domains, then place it above the Referral channel. Expect some undercounting, because links pasted into a new tab arrive without a referrer.
How often should PR teams benchmark AI answers?
Benchmark once per quarter with the same prompts, engines and scoring rubric each time. A fixed method makes the change between runs meaningful, and a quarterly schedule fits the pace at which engine answers shift.
Conclusion: Measure the Outcome, Not the Space
Every metric in this guide answers one question a CFO asks: did coverage change what buyers did or believed? Set the goal, map it to a key event or a prompt share, and report the change with the method attached.
